Procedural Shape Generation for Multi-dimensional Data Visualization

Visualization of multi-dimensional data is a challenging task. The goal is not the display of multiple data dimensions, but user comprehension of the multi-dimensional data. This paper explores several techniques for perceptually motivated procedural generation of shapes to increase the comprehension of multi-dimensional data. Our glyph-based system allows the visualization of both regular and irregular grids of volumetric data. A glyph’s location, 3D size, color, and opacity encode up to 8 attributes of scalar data per glyph. We have extended the system’s capabilities to explore shape variation as a visualization attribute. We use procedural shape generation techniques because they allow flexibility, data abstraction, and freedom from specification of detailed shapes. We have explored three procedural shape generation techniques: fractal detail generation, superquadrics, and implicit surfaces. These techniques allow from 1 to 14 additional data dimensions to be visualized using glyph shape.